Skip to content

User Guide

Comprehensive documentation for using Hyperseed's features and workflows.

Overview

This user guide covers all aspects of hyperspectral seed analysis with Hyperseed, from data preparation to advanced workflows.

Topics

Typical Workflows

Basic Workflow

graph LR
    A[Prepare Data] --> B[Run Analysis]
    B --> C[Review Plots]
    C --> D[Export Results]
  1. Ensure data is in correct format
  2. Run analysis with --export-plots
  3. Review segmentation quality
  4. Export final results

Advanced Workflow

graph LR
    A[Prepare Data] --> B[Test Config]
    B --> C{Satisfied?}
    C -->|No| D[Adjust Settings]
    D --> B
    C -->|Yes| E[Batch Process]
    E --> F[Export Results]
  1. Prepare data and create configuration
  2. Test on single dataset
  3. Iteratively adjust preprocessing and segmentation
  4. Batch process all datasets
  5. Export and analyze results

Best Practices

For Best Segmentation

  1. Use minimal preprocessing for segmentation
  2. Set appropriate min_pixels based on seed size
  3. Enable outlier removal to filter artifacts
  4. Use watershed algorithm for touching seeds
  5. Always review plots first before batch processing

For Spectral Analysis

  1. Apply appropriate preprocessing (SNV, smoothing, derivatives)
  2. Use calibrated data (not preprocessed data) for final spectra
  3. Check for bad pixels in reference images
  4. Validate wavelength calibration in header files
  5. Document preprocessing steps for reproducibility

Common Use Cases

Use Case 1: Quick Seed Count

hyperseed analyze dataset/sample_001 \
    --output results.csv \
    --min-pixels 50

Result: CSV with seed count and basic morphology

Use Case 2: Detailed Spectral Analysis

hyperseed analyze dataset/sample_001 \
    --output results.csv \
    --preprocess advanced \
    --export-plots

Result: CSV with preprocessed spectra + visualization plots

Use Case 3: Production Batch Processing

hyperseed batch dataset/ \
    --output-dir results/ \
    --config production_config.yaml

Result: Consistent analysis across all datasets

Need More Help?